Behaviors in REM sleep: AI assisted 3D Video Analysis

Rapid eye movement (REM) sleep behavior disorder (RBD) is a parasomnia characterized by abnormal jerks and dream-enacting behaviors in REM sleep. RBD, in its isolated form (iRBD), is the early stage of alpha-synucleinopathies (Parkinsons disease, dementia with Lewy bodies and multiple system atrophy), therefore its correct diagnosis will be fundamental when neuroprotective therapies will be available. Currently, iRBD can be diagnosed in specialized sleep centers with video-polysomnography (v-PSG), which requires time-consuming visual analyses. As the hallmarks of iRBD are minor dream-related motor events which might remain unnoticed, patients are often missed in the general population. This is particularly true for women, for whom the symptoms are generally mild. Furthermore, no objective outcomes are available to monitor symptomatic treatment for iRBD patients. Therefore, new technologies for automatic identification of iRBD patients and personalized follow-up in home environments are needed. 3D contactless video based on the time-of-flight principle is an emerging tool useful for this. However, the devices for recording 3D videos employed so far have never been validated against a wide range of RBD differential diagnoses, and are not appropriate for at- home and stand-alone recordings. In the context of the ERA PerMed Transnational Call 2021, the Sleep Disorder Clinic of the Department of Neurology of the Medical University of Innsbruck will coordinate a transnational European project aiming to validate personalized artificial intelligence algorithms that use 3D videos recorded with small, light and portable sensors as novel stand -alone technology for automatic identification and follow-up of iRBD patients. The new technology will be validated against gold-standard v-PSG in a total of five different European expert centers for sleep and neurology. By the end of the project, it is expected that 3D video will be validated and ready to be established as a novel and powerful stand-alone technique to reliably identify and follow-up iRBD patients. This novel tool can revolutionize the way in which iRBD patients are identified, allowing better identification of early-stage alpha-synucleinopathies. Furthermore, this technology will allow having objective measures of the efficacy of symptomatic treatments, thus making it possible to personalize them.

Rapid eye movement (REM) sleep behaviour disorder (RBD) is a sleep disorder characterised by abnormal muscle activity and dream enactment during REM sleep. In its isolated form (iRBD), it is recognised as an early marker of alpha synucleinopathies (i.e., Parkinson's disease, dementia with Lewy bodies, and multiple system atrophy). Because specialised sleep centres are few and symptoms often go unnoticed, iRBD is frequently underrecognised. The BRAVA project proposes a solution to this problem by developing and validating a novel, small, lightweight, and portable 3D video based technology that uses artificial intelligence as a powerful, automatic, stand alone tool to identify and follow up patients with iRBD. By recruiting over 100 patients with RBD and 200 controls (including patients with common differential diagnoses of RBD) across five European specialised sleep laboratories, the BRAVA project developed and validated an algorithm that detects movements during sleep with high accuracy compared with human scorers, and then automatically identifies patients with RBD using an AI based approach that leverages the rate and extent of these movements. Overall, the BRAVA project demonstrates that a small, portable depth camera can automatically detect movements and support RBD identification with performance comparable to humans, highlighting its potential as a scalable and cost effective screening aid. Future studies will evaluate the technology in home environments to ultimately validate its use for screening for early neurodegeneration.